study on circularly polarized synthetic … · proximity-coupled corner-truncated patch array...
TRANSCRIPT
-
STUDY ON CIRCULARLY POLARIZED
SYNTHETIC APERTURE RADAR: PATCH ARRAY
ANTENNAS AND SCATTERING EXPERIMENTS IN AN
ANECHOIC CHAMBER
July 2016
MOHD ZAFRI BIN BAHARUDDIN
Graduate School of Advance Integration Science
CHIBA UNIVERSITY
-
i
STUDY ON CIRCULARLY POLARIZED SYNTHETIC
APERTURE RADAR: PATCH ARRAY ANTENNAS AND
SCATTERING EXPERIMENTS IN AN ANECHOIC CHAMBER
2016 7
MOHD ZAFRI BIN BAHARUDDIN
-
ii
TABLE OF CONTENTS
Table of Contents ii
Abstract v
Abstract (Japanese) vii
List of Figures viii
List of Tables xii
Chapter 1 Introduction 1
1.1 Circularly Polarized Antennas .............................................................................. 2
1.2 Motivation and Challenges in using Circular Polarization for SAR .................... 3
1.3 Research Summary ............................................................................................... 4
Chapter 2 Synthetic Aperture Radar 6
2.1.1 Data collection Geometry ........................................................................ 8
2.2 Imaging Algorithms ............................................................................................ 10
2.3 SAR Polarizations / Radar Polarimetry .............................................................. 11
2.3.1 Linearly polarized SAR ......................................................................... 12
2.3.2 Compact-SAR ........................................................................................ 12
2.3.3 Circularly polarized SAR ....................................................................... 13
2.4 Antennas ............................................................................................................. 13
2.4.1 Antenna Design Methodology ............................................................... 13
2.4.2 Antenna Parameters ............................................................................... 14
2.4.3 Horn Antennas ....................................................................................... 17
2.4.4 Microstrip Patch Antennas ..................................................................... 18
-
iii
Chapter 3 Semi-automated SAR Testbed 20
3.1 System implementation ...................................................................................... 20
3.1.1 RF Subsystem ........................................................................................ 22
3.1.2 Positioner subsystem .............................................................................. 25
3.2 Ground test and measured data ........................................................................... 28
Chapter 4 Antenna Design 31
4.1 Reduced side-lobe Circularly Polarized Array Antenna .................................... 31
4.1.1 Single element design ............................................................................ 32
4.1.2 Array design ........................................................................................... 33
4.1.3 Array signal distribution network .......................................................... 34
4.1.4 Simulation and Measured Results .......................................................... 36
4.2 Beam tilted Circularly Polarized Array Antenna ............................................... 39
4.2.1 Single element design ............................................................................ 42
4.2.2 Steered beam design .............................................................................. 43
4.2.3 Design of reduced side-lobe ................................................................... 44
4.2.4 Array Signal Distribution Network ........................................................ 46
4.2.5 Simulated and measured results ............................................................. 48
-
iv
Chapter 5 Integrated SAR System 56
5.1 Circular Polarization Scattering Matrix .............................................................. 56
5.2 Linear Rail SAR ................................................................................................. 57
5.2.1 Single point CP range measurement ...................................................... 59
5.2.2 Range profile measurements .................................................................. 59
5.3 Rotary Turntable SAR ........................................................................................ 62
5.3.1 Radar measurement ................................................................................ 63
5.3.2 Comparison of measured results ............................................................ 67
5.4 Turntable Imaging Experiments ......................................................................... 69
5.4.1 Single Point CP Range Measurement .................................................... 70
5.4.2 ISAR Measurement ................................................................................ 73
Chapter 6 Conclusion and Future Developments 79
References 82
Appendices 87
Appendix A MATLAB codes for conditioning Raw Data .......................................... 88
Appendix B MATLAB codes for back-projection algorithm ..................................... 91
Publications 93
-
v
ABSTRACT
Observation of the Earth from aerial and space-borne platforms using synthetic
aperture radar (SAR) has provided scientists and researchers with a deeper knowledge
of the physical surface information of our planet. SAR imaging is typically performed
using linearly polarized (LP) signals and has been extensively studied. Circularly
polarized (CP) SAR imaging on the other hand is still in its early stages. Thus an
investigation into using CP signals for SAR is desired. This thesis presents the design
and testing of a SAR system using CP antennas, with an investigative focus on design
of low side-lobe array antennas. With the aim of launching a CP SAR sensor for the
purpose of remote sensing, a program to test the systems leading up to an unmanned
aerial vehicle and eventually a satellite platform was planned. All related on-board
subsystems of the SAR sensor must first be tested on the ground. In a preliminary study,
a semi-automated rail enabled ground-based SAR imaging test-bed was developed and
is able to function carrying L and C-band SAR systems. Measurements were performed
outdoors on a corner reflector and LP imaging results are presented. Following this
study, a controlled experiment was made in an anechoic chamber with the use of a
vector network analyzer (VNA) and automated rotary turntable to mount targets.
Measurements in the C-band were performed using CP signals. CP ISAR imaging
results were compared with LP ISAR images. These works will lead to the investigation
of generating images from CP receive and transmit signals, improvement of related
imaging algorithms for CP signals and suitable antenna design. This thesis continues
previous research on CP microstrip patch antennas for application in SAR
measurements. Antenna designs are based on theory and design guidelines and were
validated using electromagnetic simulation tools and measurements. Several prototype
microstrip antenna designs were studied, fabricated using traditional PCB fabrication
techniques and verified through performance parameter measurements. First, basic
patch antenna shape were studied and compared for the best performance and design
compromise based on material and fabrication method limitations. In the L-band, a
-
vi
proximity-coupled corner-truncated patch array antenna was designed. The antenna is
an L-band 1.27 GHz, 2x5 element array that has been synthesized using the Dolph-
Chebyshev method to reduce side-lobe levels. Fabricated antenna performance
characteristics show good agreement with simulated results. A set of antennas were
fabricated and then used together with the developed SAR system. Imaging results are
discussed in this thesis. In the C-band, a planar CP, beam tilted, array antenna was
studied. Several important antenna performance parameters customized to the utilized
SAR systems were achieved in this study. Among them are low axial ratio of under 3dB
with 130 MHz bandwidth via corner truncated square patches for circular polarization;
steep scanning angle of 40 degrees using phased delayed 10-element array at 0.53
lambda element spacing; suppressed side-lobe level; and reduced radiation from
spurious feeds through optimized feed network using T-junctions and mitered bends.
The optimized design was fabricated and the array antenna performance parameters
was verified through measurements in an anechoic chamber. Results show agreement
between theoretical and simulated data which match the design parameters. The
antennas and their connected SAR systems developed in this work are different from
most previous work in that this work is utilizing circular polarization as opposed to
linear polarization. To compare LP and CP results, a polarimetric analysis using
theoretical scattering matrices was made.
-
vii
ABSTRACT (JAPANESE)
SAR: synthetic aperture radar
SAR
LP: linear polarization
CP: circularly polarization SAR
CP
SAR
CP-SAR
SAR
LCSARGB-SAR:
ground-based SAR
L C
L
SAR C
LP-SAR
VNA: vector
network analyzer SAR
SARISAR: inverse SAR
CP-SAR LP-SAR
SAR
-
viii
LIST OF FIGURES
Figure 1.1: Polarizations of electromagnetic waves: (a) vertical linear polarization; (b)
circular polarization; (c) elliptical polarization. ............................................................. 2
Figure 2.1: Pulse compression radar block diagram. ..................................................... 7
Figure 2.2: Linear rail based inverse SAR measurement layout and movement scheme
in an anechoic chamber. ................................................................................................. 9
Figure 2.3: Rotary turntable based inverse SAR measurement layout and movement
scheme in an anechoic chamber. .................................................................................... 9
Figure 2.4: Methodology for designing an antenna. .................................................... 14
Figure 2.5: A linearly polarized horn antenna used to generate a circularly polarized
signal using a 90 degree phase shifter. ......................................................................... 17
Figure 3.1: SAR rail imaging platform system block diagram .................................... 21
Figure 3.2: Radiation pattern of the ETS Lindgren quad-ridged horn antenna at the
operating frequency [41]. ............................................................................................. 23
Figure 3.3: Main components of the RF Subsystem. ................................................... 24
Figure 3.4: IQ received signals from a 200 nanosecond loopback test........................ 25
Figure 3.5: Range measurement from a 200 nanosecond loopback test. ..................... 25
Figure 3.6: Positioner Subsystem layout and movement scheme. ............................... 27
Figure 3.7: The Positioner Rail SAR Imaging System mounted on modular rails used
on the test site. .............................................................................................................. 27
Figure 3.8: Arrangement of rail SAR transceiver and target corner reflector on test site.
...................................................................................................................................... 28
Figure 3.9: Range measurement with point target return of trihedral detected at a
distance of 92 meters. .................................................................................................. 29
Figure 3.10: SAR Image obtained using datasets from 381 points. ............................. 30
Figure 4.1: Effect of corner truncation size, C, on the axial ratio. ............................... 32
Figure 4.2: Theoretical, simulated and measured normalized antenna beam patterns of
the proposed array antenna. ......................................................................................... 34
-
ix
Figure 4.3: Dimensions of the proposed side-lobe reduced array antenna. ................. 35
Figure 4.4: The top and bottom layers of the LHCP fabricated array. ........................ 36
Figure 4.5: Simulated and measured reflection coefficient of the RHCP array antenna.
...................................................................................................................................... 36
Figure 4.6: Simulated and measured radiation pattern of the RHCP array antenna. ... 37
Figure 4.7: RHCP array antenna simulated and measured axial ratio and gain against
frequency. ..................................................................................................................... 37
Figure 4.8: RHCP array antenna simulated and measured axial ratio against angle. .. 38
Figure 4.9: Location of antennas and required beam pattern path. .............................. 39
Figure 4.10: Profile of desired antenna beam pattern. ................................................. 40
Figure 4.11: RF system diagram. ................................................................................. 41
Figure 4.12: Single element corner truncated patch antenna on dual substrates. Also
shown is the radiation pattern that peaks at 5 dBi at the boresight. ............................. 43
Figure 4.13: Axial ratio values as a function of the truncation size. The truncation size
is shown as a percentage of the patch width. ............................................................... 43
Figure 4.14: Comparison of a 10-element radiation pattern with and without the
proposed synthesis method. ......................................................................................... 45
Figure 4.15: Design of microstrip feed network with synthesized power distribution: (a)
Zero degree beam steer; (b) 40-degree beam steer. Note the shifted lengths of the
microstrip lines. ............................................................................................................ 46
Figure 4.16: Unsymmetrical microstrip T-junction design. ......................................... 47
Figure 4.17: Geometry of left-handed circularly polarized array antenna: (a) top view
(light gray is radiating patch, dark gray is the feed network); (b) bottom view; and (c)
side view. ..................................................................................................................... 48
Figure 4.18: Prototyped left-handed circularly polarized array antenna: (a) top
substrate; (b) bottom substrate. .................................................................................... 49
Figure 4.19: Magnitude and beam steering at different angles. 40 degrees provides the
best compromise of steer angle and gain ..................................................................... 50
Figure 4.20: Simulated 3D radiation pattern at 40 degree beam tilt: (a) decibel scale; (b)
magnitude scale. ........................................................................................................... 51
-
x
Figure 4.21: Measured and simulated return losses of the proposed array antenna. ... 52
Figure 4.22: Measured and simulated axial ratio of the proposed array antenna. ....... 52
Figure 4.23: Measured and simulated radiation pattern of the designed left-handed
circularly polarized array. Simulated maximum gain is 12.7 dBiC, measured gain is
12.1 dBiC. .................................................................................................................... 53
Figure 4.24: Proposed array module structure. From top: radiating patch, feed network,
and 3-way power divider layers. .................................................................................. 54
Figure 4.25: With a high wattage power divider, use of multiple modules can increase
the total system gain as required. ................................................................................. 54
Figure 4.26: Simulated 3D radiation pattern of a 1012 element array achieved by
utilizing three antenna modules. .................................................................................. 55
Figure 5.1: The reduced side-lobe antennas installed on the mounting on the left and
corner reflector placed on the moving platform on the far right of the image. ............ 58
Figure 5.2: Measured RL range data showing high intensity caused by reflection from
the plate on the moving platform. ................................................................................ 60
Figure 5.3: Entire raw data range bins sampled for (a) co-polarized (RR), and (b) cross-
polarized configurations (RL). ..................................................................................... 60
Figure 5.4: Range and azimuth compressed image output for (a) co-polarized (RR), and
(b) cross-polarized configurations (RL). ...................................................................... 61
Figure 5.5: Antennas and target setup in an anechoic chamber. .................................. 64
Figure 5.6: Frequency response of a steel plate at L-band 1.27 GHz with 500 MHz
bandwidth. .................................................................................................................... 65
Figure 5.7: Range profile in the time domain after transformation. ............................ 65
Figure 5.8: Range profile of the target at = 0 (perpendicular) and 90 (parallel). .. 66
Figure 5.9: Baseband ISAR (angle (), range (r)) domain data of the target. .............. 66
Figure 5.10: Conical log spiral antennas used for circular polarization (left) and flat
aluminum plate for point target measurement placed on a rotating polystyrene platform
(far right). Note the quasi-monostatic placement of the receiving and transmitting
antennas. ....................................................................................................................... 69
-
xi
Figure 5.11: Horn antennas with phase shifters used for circular polarization (right) and
a dihedral reflector for point target measurement placed on a rotating polystyrene
platform (far left). Note the quasi-monostatic placement of the receive and transmit
antennas. ....................................................................................................................... 70
Figure 5.12: CP frequency response of a flat plate at C-band 4-8 GHz with background
noise removed. Note the relatively high readings for cross-polarized setup compared to
co-polarization. ............................................................................................................ 71
Figure 5.13: Range profile of RL and RR data for perpendicular flat plate after IFFT
range compression. ...................................................................................................... 72
Figure 5.14: Phase difference of the co-polarized scattering signals detected from a
dihedral reflector is close to 180 degrees. .................................................................... 72
Figure 5.15: Comparison of linear (LP) and circular polarized (CP) imaging output of
a flat plate reflector. ..................................................................................................... 74
Figure 5.16: Comparison of LP and CP imaging output of a dihedral reflector. ......... 75
Figure 5.17: Comparison of LP and CP imaging output of three thin vertical poles. . 76
Figure 5.18: A model airplane used a complex target to be imaged by the developed
system. ......................................................................................................................... 77
Figure 5.19: Comparison of LP and CP imaging output of a model airplane. ............ 78
-
xii
LIST OF TABLES
Table 3.1: SAR imaging system design specifications ................................................ 21
Table 4.1: Derived Dolph-Chebyshev Array Factors for 2x5 Planar Array ................ 34
Table 4.2: Proposed Array Antenna Dimensions ........................................................ 35
Table 4.3: Circular polarization (left/right-handed) notation for receive and transmit
antennas. ....................................................................................................................... 41
Table 4.4: Derived excitation coefficients and phase delay for each patch element. .. 45
Table 5.1: Linear and circular polarization basis scattering matrices for canonical
targets used in experiments. ......................................................................................... 57
-
1
Chapter 1 INTRODUCTION
Polarization of antennas was first studied by Heinrich Hertz around 1886-1889 when
he demonstrated that a receiver does not respond when the dipole antennas are aligned
perpendicular to one another. He further showed that as the receiver is rotated the
received signal increases and reaches a maximum strength when the dipole antennas
are parallel [1]. Polarization of waves and antennas can be classified into three
categories: linear, circular, and elliptical polarizations. Elliptical polarization is a
general term and radio systems are rarely designed for use with this polarization
specifically, even though linear and circular polarizations are the extreme special cases
of elliptical polarization. Instead radio systems are typically designed to use linear or
circular polarizations.
Linear polarization (LP) is easily generated and is commonly used for mobile
phones, communications broadcasting, and automotive radar, among other classical
radio applications. A pure LP signal should have only one electric field component,
either horizontal or vertical. Thus, LP signals can be further classified as horizontal and
vertical polarizations. Circular polarization (CP) on the other hand is used for global
navigation satellite systems, near-field RFID and weather radar. CP also has two sub-
classifications, left-handed and right-handed circularly polarized, depending on the
rotation direction of the electric field vector [2]. A graphical representation of each
polarization showing the electric field components is shown in Figure 1.1. The selection
of polarization is based on the application as there are some characteristics of a
polarized waveform that can be more advantageous for specific applications.
-
2
(a) (b) (c)
Figure 1.1: Polarizations of electromagnetic waves: (a) vertical linear polarization; (b)
circular polarization; (c) elliptical polarization.
1.1 Circular Polarization
To generate circularly polarized radio signals, two orthogonal electric field
components that are equal in amplitude and 90 degree phase shifted are required.
Designing pure CP antennas without the use of external hardware such as a phase shifter
is challenging. However, CP signals offer several benefits over LP signals:
1. Immune to Faraday rotation. As electromagnetic waves travel through the
Earths atmosphere, it experiences a rotation of the plane of polarization. This
rotation occurs more in the upper atmosphere where the travelling wave is
affected by ionized plasma that generates strong magnetic fields. Studies have
shown that the Earths magnetic field varies in strength throughout the globe,
and depends on many other factors such as time of year and solar activity [3].
LP signals will be affected by this Faraday rotation and will experience
polarization mismatch. Circular polarization however is immune to this effect
because both orthogonal components have equal magnitude and are rotated by
the same angle. This is the main reasoning for CP to be used in most Earth-to-
space satellite communication systems [4].
Propagation direction
Vertical E-field
onlyHorizontal
E-field Vertical E-field
Equal E-field
magnitude
Unequal E-field
magnitude
90 phase
difference
-
3
2. No multipath propagation. Circularly polarized waves that are reflected by
flat conductive surfaces will become counter polarized. For example, a right-
handed CP wave will be reflected as a left-handed CP wave and vice-versa. This
inherent property of the CP reflection allows the antenna to filter out reflected
signals and is most beneficial to satellite based navigation systems [5]. CP is
also advantageous in indoor based high data rate communications as it decreases
interference between direct and reflected signals.
3. No polarization loss due to misalignment. Linearly polarized antennas must
be precisely aligned to avoid polarization mismatch. CP antennas are not
affected by polarization mismatch. Additionally, LP antennas can still receive
pure CP signals with a loss of 3dB. This property is used in RFID tagging
systems to ensure tags are detectable regardless of orientation.
1.2 Motivation and Challenges in using Circular Polarization for SAR
Circularly polarized antennas are to be designed to take advantage of the
benefits stated in the previous section. Typical features of LP antennas such as
wideband and pure polarization are not immediately available to CP antennas. The aim
of this work is to propose antenna designs for use in an airborne or spaceborne SAR
system. The antennas need to comply with designed SAR performance requirements,
be circularly polarized, low mass, low profile, and include a beam that is shaped to
reduce side-lobes and steered to a specified look angle. The designs are to be
implemented in a synthetic aperture radar system.
Using an existing L-band pulse Doppler SAR measurement system [6], an initial
pilot study was performed [7]. L-band has large wavelength which results in systems
and equipment to be physically large. Measurements is only feasible in an outdoor range
with large bulky reflector targets being placed at least 100 meters in range and a
platform movement of at least 40 meters. Even though the pilot study was a success,
the time consumption, man-power, equipment, and outdoor RF measurement licensing
requirements [8] forced the study to be moved indoors.
This led to the development of an indoor reconfigurable wide-band vector
network analyzer (VNA) based SAR system for testing antennas in different bands [9].
-
4
The setup was suitable for frequencies above 2 GHz. Lower frequencies are still
possible but it is close range, almost near field. Several standard and custom designed
LP and CP antennas were used to make SAR measurements in the indoor range. As
opposed to the pulsed radar L-band system, this indoor VNA based system generates
signals that are stepped frequency continuous wave (SFCW). SAR imaging with SFCW
requires comparatively larger bandwidth to provide the same resolution.
To compare the LP and CP SAR imaging output, an understanding of
polarimetric scattering is required. The scattering matrices for different target types are
completely different between the two polarizations. An additional challenge imposed
when performing polarimetric measurements is polarimetric calibration. LP calibration
is well established but CP calibration is relatively unexplored. Calibration is an
important step in the post-processing of the captured scattering results but is outside the
scope of this thesis.
1.3 Research Summary
A synthetic aperture radar imaging system for studying imaging with circularly
polarized antennas is developed in this thesis. This system is reconfigurable for imaging
using a linear rail-based SAR or a turntable-based inverse SAR. This system uses a
combination of range-gated stepped frequency continuous wave (SFCW) using a VNA
as the transceiver, circularly polarized patch array antennas customized to specific
frequency bands, and a back-projection algorithm (BPA) to produce the SAR image.
The innovation in this thesis is mainly focused on developing circularly polarized array
antennas as part of a new synthetic aperture radar system. Unlike previously developed
systems which receive, transmit and process LP signals, the developed antennas and
SAR measurement system were specifically designed for the study of CP signals for
imaging from airborne or spaceborne platforms.
Several research topics were developed to accomplish this. Chapter 2 provides
a brief background on SAR data collection geometry, synthetic imaging algorithms,
state-of-the-art polarimetric SAR systems, followed by proposed innovations in
antenna array design. In Chapter 3, an automated rail-based SAR measurement system
was developed and tested in the L-band using linear polarization. After proving the
-
5
functionality of the measurement system, Chapter 4 presents development of CP array
antennas with reduced side-lobes to improve cross-range ambiguity. Beam tilting is also
explored to conform to vertical mounting requirements of a targeted airborne platform
installation. Finally, the CP antenna and SAR imaging system is integrated and CP
measurements were performed in an anechoic chamber as presented in Chapter 5. It
was found that CP imaging results are in agreement with theoretical analysis. Future
work and conclusions are discussed in Chapter 6.
-
6
Chapter 2 SYNTHETIC APERTURE RADAR
An application utilizing radar systems that transmit a high power radio frequency signal
and receives the backscattered echoes sequentially and then using decompression
techniques to generate a high resolution image of the radar target is known as synthetic
aperture radar (SAR). The SAR sensor itself is an imaging radar mounted on a moving
platform. Electromagnetic waves are sequentially transmitted and the backscattered
echoes are collected by the radar antenna. An appropriate number of received signals
along the moving path allows the construction of a synthetic aperture that is much larger
than the physical antenna length. The image that results after processing the raw data
represents a measure of the scene reflectivity [10]. SAR has grown quickly over the last
two decades to become one of the main instruments for remote sensing from airborne
and space-borne platforms. Its inherent characteristics that provide high resolution
images, all weather sensing and day and night monitoring have made it invaluable for
remote sensing disaster monitoring applications [11] [12]. Past space-borne SAR
systems are typically installed on relatively large satellites in the range of 1,000 to 4,000
kilograms such as the TerraSAR-X, ALOS and ALOS-2. Heavier payloads will require
larger rockets with greater thrusts, which will inevitably be expensive to launch into
orbit. Smaller and lighter satellites are cheaper to launch and can even be launched with
other payloads as a piggyback [13].
With this in mind, the Josaphat Microwave Remote Sensing Laboratory
(JMRSL) has embarked on a project to develop a unique space-borne SAR sensor with
the aim of implementing the entire SAR system in a compact sized microsatellite. The
proposed system is envisioned to be within 50 x 50 x 70 centimeters in size when
stowed and weighing less than 100 kilograms [14]. Among the basic research planned
-
7
to be conducted from the satellite is mapping of land surface, interferometry, and wave
scattering mechanism of vegetation, snow or ice, rocks and desert areas. Potential
applied research includes ocean monitoring, snow or ice monitoring and disaster
monitoring.
Spaceborne SAR missions mostly operate in linear polarization (HH, VV, HV
and VH) [4]. When linear polarized signals propagate through the Earths atmosphere
it can be sensitive to the Faraday rotation effect, particularly at low radio frequencies
bands such as L-band. Our work focuses on a proposed circularly polarized synthetic
aperture radar (CP-SAR) to obtain land deformation and physical information of the
Earths surface [14] [15].
In preparation for the spaceborne SAR imaging platform, an unmanned aerial
vehicle (UAV) and an initial ground-based test-bed has been developed to evaluate
outdoor performance of the required radio frequency (RF) transceiver, chirp-pulse
generator and antenna subsystems [15]. The developed radar system uses a pulse
compression signal processing technique as shown in the block diagram in Figure 2.1.
Figure 2.1: Pulse compression radar block diagram.
To test our antennas as well as imaging algorithms in a controlled environment,
we have developed a simple SAR system for use in an anechoic chamber with a vector
network analyzer (VNA) as the signal generator and capturing device. Already having
a rotary turntable installed in our anechoic chamber, an inverse synthetic aperture radar
(ISAR) configuration was adopted. A developed linear rail system was also used in
experiments for comparison with the turntable results. As opposed to SAR, inverse
SAR is an imaging method that utilizes relative motion of a target with respect to a
stationary transceiver radar to generate an image of the target. As the radar beam
Modulator
/ ramp
generator
Video amplifier (LPF)
Video out
Mixer
Splitter
LNA
PA
Osc
TX
RX
-
8
remains fixed and while the target rotates, reflectivity data from different reflectors on
the target can be measured.
Previous works on ISAR have introduced fundamental concepts for data
collection and processing [16], studied on improving processing speeds by using Fast
Fourier Transforms (FFT) instead of Discrete-Time Fourier Transform (DTFT) [17], or
performed radar cross section measurements (RCS) using an ISAR configuration in an
indoor measurement range [18]. The mentioned works included core concepts of ISAR
as well as proposed techniques to improve image quality. The systems developed in this
work are different from most previous work in that this work seeks to use circularly
polarized (CP) radiation signals to generate images. This work will lead to the
investigation of CP images, improvement of related imaging algorithms for CP signals
and suitable antenna design.
2.1 Data collection Geometry
Inverse synthetic aperture radar is an imaging mode that takes into account the
relative motion of a target with respect to a stationary transmitting/receiving antenna to
image that said target [17]. To generate an ISAR image of a target from the returned
signals two conditions need to be satisfied; (1) backscattered data has to be in two-
dimensional format, (2) radar imaging geometry can provide returned data that contains
slightly different geometrical information about the target. To comply with the two-
dimensional format requirement, a frequency varying wave was used to obtain the range
resolution for the first dimension. Two approaches were made to meet the second
dimensional conditions: a rotary turntable for a rotational axis, and a linear rail for a
Cartesian axis.
The measured backscattered signal can be interpreted as the radar cross section
(RCS) of the target at the incident angle over the entire measurement bandwidth. There
are various RCS measurements depending on the type of signal being transmitted [18].
We chose to measure the scattering data using a stepped frequency continuous wave
(SFCW) in our measurements. In SFCW measurement, a band of frequencies are
transmitted. This carries more information in terms of frequency components as
compared to a single frequency element. Collected range response data is processed
-
9
using matched filtering and Fast Fourier Transform to convert from frequency domain
into time domain, and thus obtaining a measureable distance range profile [17] [19]. To
generate the ISAR image, the data for each pixel is a two-dimensional inverse Fourier
transform of range profiles of all the viewing angles [20] [21]. Reflection measurement
geometry for the linear rail and rotary turntable based measurement is shown in Figure
2.2 and Figure 2.3 respectively.
Figure 2.2: Linear rail based inverse SAR measurement layout and movement scheme
in an anechoic chamber.
Figure 2.3: Rotary turntable based inverse SAR measurement layout and movement
scheme in an anechoic chamber.
Transmit
antenna
Receive
antenna
Range, Rm
Anechoic chamber
RF
Transceiver
FPGA
Chirp
Generator
ADC/
DAC Computer
Rail robot
controller
Approximate
beam pattern
Pla
tfo
rm m
ov
emen
t
alo
ng c
ross
-ran
ge
len
gth
, L
sa
Platform
Control
tether
Reflector
Transmit
antenna
Receive
antenna
Range, Rm
Anechoic chamber
Agilent
E8364C
VNA
Computer
Turntable
motion
controller
Approximate
beam pattern
Control
tether
Target
Port 2
Port 1
Turntable
rotation
m
GPIB connection
-
10
For a rotary turntable configuration in Figure 2.3, the reflection from the target
can be considered as composed of multiple independent point reflectors that is
proportionate to the scatterer reflectivity function with the assumption that the
antennas radiation pattern is the same for all measured points in a cylindrical
configuration [24]. The reflectivity function is measured in the cylindrical coordinate
(,), and then projected into the Cartesian image plane (x,y). The scattering
parameters measured at a specified angle m for a point scatterer target arbitrarily
positioned at (x0,y0) or (0,0) can be expressed as:
mrm
r RjkR
kSm
exp1
,200
(2.1)
where kr is the radiating signal two-way propagation wavenumber defined as kr = 4f/c,
and Rm is the measured range distance from the antenna to the target. For expressing
the scattering parameters viewed from every possible angle, Eq. (2.1) can simply be
integrated over the entire view range as:
ddRjkR
kS mrm
rm
2
0 0 200 exp
1, (2.2)
2.2 Imaging Algorithms
In this work the back projection algorithm was utilized to process raw measured
data for the image generation. The algorithm was made famous by [22] and has the
advantage of being able to reconstruct images even when data is incomplete or not
perfectly aligned to the measurement path. This makes it suitable for airborne missions
where the flight path may shift or the aircraft body rolls slightly [23]. There has also
been application of the back projection algorithm in prototype security systems to detect
trespassers or concealed objects [24].
Back projection works by taking a number of finite projections of an object
which is measured from the radar system. The projections are measured from different
angles or views, each generating an image. The basic function of the algorithm is to
simply run the projections back through the image, thus the name back-projection, to
generate a rough approximation of the target object. These projections will add
-
11
constructively in areas that have multiple scatterings. Projection reconstructed images
will usually result in blurred artifacts surrounding the object, which can be removed by
passing the image through a high pass filter.
The reflectivity image in the Cartesian plane is given by the inverse Fourier
transform (IFT) of the reflectivity function, (x,y), as [24]:
mrrmrr ddkkRjkkSyx m
0exp, (2.3)
The integral inside the bracket of Eq. (2.3) can be viewed as a one-dimensional IFT as
a function () = () , evaluated at a specified range, Rm. By defining
() as the IFT, Eq. (2.3) is simplified to:
mm dRqyx m
, (2.4)
Eq. (2.4) is the back projection algorithm that is used in this study to generate Cartesian
images from reflection measurements of targets on a rotating turntable. Implementation
codes in MATLAB are presented in the Appendix.
2.3 SAR Polarizations / Radar Polarimetry
The polarization of radiation is an important property when discussing
propagation and scattering of electromagnetic waves. Polarization refers to the electric
field vector locus in the plane perpendicular to the propagation direction. The length of
the vector denotes the wave amplitude and the rotation rate of the vector denotes the
frequency of the wave. Polarization then refers to the shape and orientation of the
pattern that is traced by the vector. Figure 1.1 visualizes this showing in (a) linear
polarization where there is only a single electric field vector component that is either
vertical or horizontal, (b) circular polarization where there are two orthogonal electric
fields of equal amplitude that have a 90 degree phase shift between them, and (c)
elliptical polarization where the amplitude and phase can be different between two
orthogonal electric fields.
Depending on the property of the target, the transmitted wave can change after
reflection and the backscattered wave can be in the form of different polarizations. The
-
12
study and analysis of these receive and transmit polarization combinations is the science
of radar polarimetry [26] [27].
2.3.1 Linearly polarized SAR
Most SAR radars are designed to transmit radiation that is either vertically
polarized (V) or horizontally polarized (H). This is because polarization on either
receiving or transmitting signals can be synthesized with distinct relationships between
them. In this thesis, all receive-transmit pairs are ordered by the receive polarization,
followed by the transmit polarization. There are four typical combinations of receive
and transmit polarizations:
HH - horizontal transmit and horizontal receive
VV - vertical transmit and vertical receive
VH - vertical receive and horizontal transmit, and
HV - horizontal receive and vertical transmit.
HH and VV polarization combination is also referred to as co-polarized because the
receiving and transmitting polarizations are the same. Likewise, the VH and HV
combinations are referred to as cross-polarized because the receiving and
transmitting polarizations are orthogonal to each other.
Polarimetric radar can be utilized to the find the scattering matrix or target
response using two orthogonal polarizations. Given a known scattering matrix of a
target, computation of any combination of incident and received polarization is possible.
Through this polarimetric combination, the detectability of selected features in an
image can be enhanced. This is where full polarimetric measurement is most powerful.
2.3.2 Compact-SAR
Compact polarimetry is a recent technique using dual-polarization SAR systems
to construct quad-pol information. Compact polarimetry shows advantages of being
able to reduce size, complexity, price and resolution of a SAR system while possibly
maintaining the same capabilities of a full polarimetric system. First introduced by
-
13
Souyris et al. [28], the original setup uses a linearly polarized transmit antenna mounted
45 degrees with respect to the horizontal and vertical receive antennas.
Follow up research has included comparative studies of different Compact-SAR
modes [29] such as those using circular polarization for transmit, and polarimetric
calibration techniques [30]. Adoption of the Compact-SAR in actual application are
being tested in the NASA-ISRO SAR Mission (NISAR) [31] and JAXAs ALOS-2
experimental mode [32].
2.3.3 Circularly polarized SAR
The most recent mode in polarimetric construction of SAR images is the
circularly polarized SAR (CP-SAR). In this mode, both receive and transmit antennas
are circularly polarized (CP). Initial concepts have shown that polarimetric
combinations in CP provide different backscattering information than that of linearly
polarized systems [33] [34] [35] [36]. This can provide further insight into properties
of observed targets. It is the aim of this work to contribute to the realization of the fully
polarimetric circularly polarized SAR system.
2.4 Antennas
2.4.1 Antenna Design Methodology
The antenna design methodology is shown in the flowchart of Figure 2.4.
Designing an antenna begins with considering the designed SAR requirements and
combined with a new antenna concept. After drawing and designing the antenna, it is
simulated in the FEKO full wave simulation tool. If the simulated performance
parameters achieves the expected performance, the design is taken for fabrication.
Prototype is fabricated in-house using traditional chemical etching process. The
completed prototype is measured using an Agilent E8364C PNA Network Analyzer for
the reflection coefficients. Far field radiation parameters such as the radiation pattern,
gain and axial ratio are measured in the JMRSL anechoic chamber using the same
network analyzer and an ETS Lindgren 3102 conical log spiral antenna. The chamber
is installed with 45 cm long pyramidal carbon loaded urethane foam absorbers to
-
14
suppress reflections. A rotary turntable in the chamber allows measurements to be
performed in a single dimension. Measurement setup and calculations are made based
on techniques recommended by the IEEE Std 149-1979 (R2008) Standard Test
Procedures for Antennas [37].
Figure 2.4: Methodology for designing an antenna.
2.4.2 Antenna Parameters
Antenna performance is evaluated based on many parameters. The antenna that
is being tested for performance is known as the antenna under test (AUT). Generally
antennas are designed specifically for each application, thus the importance of each
parameter varies depending on its use. In the case of CP-SAR requirements, the
following performance parameters are measured.
2.4.2.1 Reflection coefficient
Reflection coefficient is the amount of power that is reflected by a port i,
referred to as Sii measured in dB. According to the law of conservation of energy, any
Concept
Design
Simulation in
FEKO
Performance
requirements
Prototype
Measurement
Performance
satisfied?
Performance
satisfied?
Disseminate
Yes
Yes
No
No
-
15
power that is not reflected gets converted into heat, or radiated by the antenna. Typically
an antenna has a single port; port i = 1. Ideally an antenna port should reflect zero power,
effectively radiating all the power. An antenna is considered good at radiating at a
certain frequency, f, if the S11 at that frequency is below 10 dB. The range of
frequencies where the criteria S11(f) < 10 dB is fulfilled is called the impedance
bandwidth. The S11 is measured directly at the antenna port using a vector network
analyzer.
2.4.2.2 Radiation pattern and gain
It is very important to know the radiation direction of an antenna. This can be
known from the antennas radiation pattern and gain. The radiation pattern shows the
power radiated by the antenna at all angles as a function of and with the center of
the antenna as the origin. The power values are normalized to its maximum value to
provide information on the antenna angular properties. The radiation pattern alone,
however, does not provide information on the exact power being radiated.
Directivity is a measure of how directional an antenna is. An antenna with zero
directionality radiates equally in all directions (isotropic), thus this antenna would have
a directivity of 1 (or 0 dB). Conversely an antenna that is very directional will have a
high directivity value. In remote sensing, such as our application of antennas in SAR,
would require an antenna with relatively high directivity to maximize power transfer
and to reduce signals from undesirable directions. Directivity of an AUT can be defined
as ratio of radiation intensity in a specified direction to the radiation intensity of an
isotropic antenna. The radiation intensity of an isotropic antenna, Uiso, can be written
as:
4
, radisoP
U (2.5)
where Prad is the total radiated power.
Using this with the directivity definition, an AUT would have a directivity, D, of:
rad
AUT
iso
AUT
P
U
U
UD
,4
,
, (2.6)
-
16
where UAUT(,) is the radiation intensity at a specified direction. However, it is very
difficult to practically measure the total radiated power. Thus the definitions of gain
and realized gain are used as acceptable performance parameters.
The gain of an antenna describes how much power is radiated in the peak
radiation direction compared to that of an isotropic antenna, which should be all the
power that is not reflected at the input port. The gain, G, in a specified direction can be
calculated as:
in
AUT
P
UG
,4 (2.7)
where Pin is the electrical power received by the AUT. Gain can also be defined as
directivity multiplied by the radiation efficiency, which in turn is a ratio of total radiated
power over input power. In practical measurements, the gain of the AUT is the ratio of
measured power in a given direction to the measured power of a reference antenna in a
reference direction. The gain value is dimensionless and express in decibel isotropic
(dBi). Dipole antennas are used as reference antennas to calculate the gain. A set of left
and right-handed circularly polarized conical spiral log antennas were used as gain
standard antennas for their high degree of polarization purity. Gains measured with
reference to these circularly polarized antennas are expressed in dBiC.
2.4.2.3 Axial ratio
Axial ratio (AR) is a measurement of polarization purity of the AUT to be
circularly polarized. Circularly polarized signals consists of two orthogonal E-field
components of equal magnitude that are 90 degrees out of phase. The axial ratio is then
the ratio of the two E-field components. A pure circularly polarized wave has an AR of
1 (or 0 dB), whereas a linearly polarized wave has an AR = . In CP antenna
measurement, AR is the cross polarization ratio between LHCP and RHCP radiation
intensity in a given direction, given by the following equation [38]:
20
20
10
101
101log20
LHCPRHCP
LHCPRHCP
UU
UU
dBAR (2.4)
-
17
where URHCP and ULHCP are the measured radiation intensities in dB for a specified
angle and frequency with respect to the reference RHCP and LHCP antennas.
The axial ratio bandwidth is the frequency range where the AR value is under a
specified value. In this study, an antenna is considered to be circularly polarized when
it has an AR of under 3-dB. The radiation pattern can be given by an amplitude, phase,
and polarization at each direction of space.
2.4.3 Horn Antennas
In terms of antenna design, it is possible to modify the feed method of an LP
antenna for it to generate CP signals. For example an LP quad-ridged horn antenna can
have two input ports, one for horizontal and another for vertical polarization. If a signal
is input into a 90 degree phase shift splitter, the in-phase output and 90 degree phase
shifted output can together be fed into the LP horn antenna to generate a CP signal. An
example of this is shown in the circuit and image of Figure 2.5.
Figure 2.5: A linearly polarized horn antenna used to generate a circularly polarized
signal using a 90 degree phase shifter.
90 phase shift
power divider
90
0
Input
-
18
2.4.4 Microstrip Patch Antennas
Microstrip patch antenna technology has matured since it was introduced in the
year 1953. It is now widely used in mobile phones, aircraft radar, interplanetary
communications, and satellites, where size, low profile, weight, performance and ease
of installation are requirements. They are relatively simple and inexpensive to
manufacture using traditional printed circuit techniques. When mounted on rigid bodies,
they are very robust and can conform to different shapes. Patch antennas have some
drawbacks however, in terms of low efficiency, low power, low polarization purity,
spurious feed radiation and very narrow frequency impedance bandwidth [2]. Over the
years, researchers have improved patch antenna designs to overcome the said
drawbacks but balancing the performance with design complexity is an engineering
challenge.
2.4.4.1 Feeding techniques
The antennas required in this study need to be circularly polarized. CP can be
achieved by exciting two orthogonal modes with a 90 degree time-phase difference
between them. For CP enabled patch antennas, there exist several prominent techniques.
For a square shaped element, circular polarization can be generated by simply feeding
the patch at two adjacent sides with signals that are 90 degree phase shifted from each
other. This method adds complexity to the feed and can be difficult to design especially
for large arrays. The two feed method can also be used in coax fed circular patches.
Since the feed network for multiple feed can be quite challenging and contribute
to spurious radiation, a single feed is preferred. For a single feed, there are designs to
excite the patch diagonally from a corner of a nearly square patch. Rectangular patches
can also be fed along the diagonal line to create circular polarization. These designs
will result in a very narrow bandwidth.
2.4.4.2 Truncated corner
Another popular single feed technique is to generate CP signals by truncating
two opposite corners of a square patch. The truncation will cause rotation of surface
waves on the patch which in turn will radiate circularly polarized signals. This method
-
19
has the advantage of simple feed network, highly reduced unwanted spurious radiation,
and wider axial ratio bandwidth compared to the diagonal feed.
2.4.4.3 Spiral antennas
Spiral antennas are used in ultra-wideband (UWB) applications. These antennas
are defined by angles, such that they are also referred to as frequency independent
antennas. They are however significantly larger in size and the beam pattern changes
depending on the input frequency. The antenna generates an RHCP beam on one side
and an opposite LHCP on the reverse side. Since the frequency range is very wide,
reflectors cannot be used to block the rear lobe. Reflectors need to be placed 0/4 away
and would thus make it frequency dependent. Absorbers can be used on the other hand
but this will reduce the efficiency by half.
-
20
Chapter 3 SEMI-AUTOMATED SAR TESTBED
An initial ground-based test-bed was developed to evaluate the performance of the
required radio frequency (RF) transceiver, chirp-pulse generator and antenna
subsystems in a pilot study. The aim of the test-bed is to safely perform ground tests on
all the required SAR and flight control subsystems, while also finding the best setting
to provide the fastest scan time along the azimuth or cross-range length using the rails.
The rail imaging platform and its systems developed in this work are different from
most previous work [39] [40] in that this work is in the L-band, which entails relatively
larger sized hardware due to the longer wavelength. This chapter presents the RF
subsystem, which generates, transmits and receives the electromagnetic waves at the
front-end and also the developed rail-based Positioner Subsystem moving platform.
3.1 System implementation
The developed rail-based SAR imaging platform simulates the linear movement
of an airplane or satellite and forms the aperture required for a ground-based SAR
system. As shown in Figure 3.1, our ground-based SAR is divided into (1) the RF
Subsystem, which generates, transmits and receives the electromagnetic waves; and (2)
the Positioner Subsystem, which manages all movement of the rail platform. All
components under the box marked Positioner Moving Platform are mounted on the
rail and are moved all at once. Platform movement controls are managed by equipment
at the control station. A technical specifications summary of the SAR system is shown
in Table 3.1
-
21
Figure 3.1: SAR rail imaging platform system block diagram
Table 3.1: SAR imaging system design specifications
No. Rail SAR Technical Specifications
Specifications Design Value Remarks
1. Operating frequency 1.27 GHz L-band,
= 23.6cm
2. Bandwidth 100 MHz
Resolution:
1 meter (azimuth),
1 meter (range)
3. Waveform, pulse width FM Pulse 0.5s pulse width
4. Polarization Single, VV VV transmit and
receive
5. Antenna Aperture 30 cm Quad-ridged horn
antenna
6. Antenna Gain 20.4 dB 47 dBm
transmit power
7. 3dB beamwidth 30.0
Azimuth range
illumination:
~200 m at range
distance 1000 m
8. Signal-to-noise ratio (SNR) > 10 dB at 1000 m
9. Overall Sensor Weight < 35 kg
Positioner
Controller
Motor
driver
Platform
motors
FPGA
DAC/ADC
RF Subsystem
Embedded PC
Control Station
PC
RF Transceiver
RF Subsystem
Positioner Subsystem
Tx Antenna
Rx AntennaP
osi
tio
ner
Mo
vin
g P
latf
orm
Co
ntr
ol S
tati
on
-
22
3.1.1 RF Subsystem
The RF Subsystem consists of the RF Transceiver embedded computer which
saves raw data and controls the FPGA. The FPGA generates the chirp pulse and triggers
the RF Transceiver to transmit and receive the radar pulse. The digital-analog converter
(DAC/ADC) handles signal con-version between the FPGA and RF Transceiver. Our
customized RF Transceiver takes the chirp-pulse and runs it through a high power
amplifier (HPA) to be transmitted by the antenna. Two open boundary quad-ridged horn
antennas were used for their good gain and low VSWR in the L-band. Two antennas
are used; one for transmit, another for receive. Both antennas are set to vertical
polarization (VV).
The system uses a pulse-Doppler radar architecture. Our systems operating
center frequency is in the L-band because this frequency band is suitable for deep
penetration of media. Among the planned observation targets are vegetation, snow and
soil. The RF transceiver is designed with a rated transmit output power of 50 Watts.
Sizing for the system begins with calculations suggested by [1]. The implemented
system operates at 1.27 GHz with a bandwidth of 100 MHz. The horn antenna has a 30
centimeter diameter da and thus an azimuth antenna beamwidth of
rad 786.03.0
236.0
a
ad
(3.1)
where is the wavelength. The slant range resolution r is inversely proportional to the
system bandwidth according to r = c0/2Br, where c0 is the speed of light. This is
followed by the synthetic aperture length given by Lsa = r0/da = 70.81 meters and
azimuth resolution:
meters 15.022
0 a
sa
a
d
Lr
(3.2)
This resolution can be achieved using multi-pass techniques. Thus, for single pass scans
we have assumed an image resolution of 1 meter. Figure 3.2 shows the radiation pattern
of the horn antenna employed in this SAR ground measurement [41]. It can be seen in
the figure that the 3dB beamwidth of the horn antenna is approximately 30.0.
-
23
Figure 3.2: Radiation pattern of the ETS Lindgren quad-ridged horn antenna at the
operating frequency [41].
The radar system front-end is the RF Transceiver module and can be seen in the
block diagram of Figure 3.3. LO is the local oscillator generating the 1.27 GHz
sinusoidal continuous waveform. The LO signal is modulated with the chirp signal
coming from the digital-to-analog converter (DAC). Band-pass filters (BPF) are used
to reduce noise distributed by nearby components. The modulated signal is fed through
a high-power amplifier (HPA) to achieve the pulse transmission output power of 47
dBm. The amplified signal is then fed through a 3-port circulator with a 50 matched
load on one of the circulator ports. This effectively makes it an isolator to block any
unwanted return signals that may come from the antennas. The chirp signal from the
circulator is then radiated towards the target scene via the transmit horn antenna.
-
24
Figure 3.3: Main components of the RF Subsystem.
Reflected signals from the target scene are then received by the receive antenna,
passed through a circulator and limiter. Since return signals are typically much smaller
than the transmit signal, signals from the receive antenna are fed through a low-noise
amplifier (LNA) with 30 dB gain. The output of the LNA is fed through a band-pass
filter and continues into the RF port of the mixer in the I/Q de-modulator. The
intermediate frequency (IF) output of the mixer is passed through an in-phase (I) and
quadrature-phase (Q) demodulator. The IQ signals are then digitized using a pair of
analog to digital converters (ADC). These IQ signals contain the frequencies which
provide the range to target information.
To test the RF Subsystem, the RF chirp-pulse amplified output is looped back
through 200 and 100 nanosecond delay lines and attenuators back into the transceiver
input. The respective I and Q signals for the 200 nanosecond delay line loopback test
are shown in Figure 3.4. After performing inverse fast Fourier transform (IFFT) on the
IQ signals, the range measurement shows a peak at 30 meters as shown in Figure 3.5.
This corresponds to the length of the delay line being used.
Analog/Digital
Converters
DAC
RF Transceiver
Field Programmable Gate Array
Embedded
RAM memory
BPF BPF
HPA
LNA
I/Q Demodulator
ADC
Chirp
generator
BPF
LO
Switch
ADC
RF Subsystem Embedded PC
FPGA
Manager
SSD
MemoryUser
Interface
Switch
Tx Antenna
Rx Antenna
-
25
Figure 3.4: IQ received signals from a 200 nanosecond loopback test.
Figure 3.5: Range measurement from a 200 nanosecond loopback test.
3.1.2 Positioner subsystem
The RF Subsystem is carried by the moving platform controlled by the Positioner
Subsystem. The antenna Positioner Subsystem ensures the moving platform is
positioned according to the movement interval requirements along the cross range. All
SAR subsystems were placed on the Positioner moving platform to reduce RF power
losses and signal delays since the cross range distance of 50 meters would have required
coaxial cables that are far too long.
0 500 1000 1500 2000 2500 3000 3500-0.5
0
0.5
0 500 1000 1500 2000 2500 3000 3500-0.5
0
0.5
Samples
Am
plitu
de,
Q-c
han
nel
Samples
Am
plitu
de,
I-c
han
nel
0 50 100 150 200 250 300 350 400 450 500-80
-70
-60
-50
-40
-30
-20
-10
0
10
Range, m
Rela
tive P
ow
er,
dB
Distance in meters (m)
Rel
ativ
e po
wer
in d
ecib
els
(dB
)
Corresponding spike after 200 ns delay
-
26
Due to the criticality for the raw data to be as precise as possible, the Positioner
is designed to have a precision of 0.1mm. The ramp up and ramp down speeds have
been tuned to minimize vibrations that would otherwise be apparent due to the large
inertia caused by the relatively large mass of the system. A complete system installation
with the Positioner mast, two horn antennas, RF transceiver, chirp generator and data
collector add up to a total of 30kg of mass.
Movement sequence of the platform is managed by a user interface on the
Control Station PC where the user inputs parameters such as total distance, interval
length and motor speed. The user interface sends speed and position commands to the
motor Positioner Controller module via a General Purpose Interface Bus (GPIB). The
Positioner Controller then controls the rotation speed of the motor and onboard
encoders provide rotational position feedback.
For each interval the Control Station PC moves the platform to the desired
location, and then sends a command via network socket to the RF Subsystem Embedded
PC, which in turn controls the radar to transmit a chirp-pulse signal. Once the RF
Subsystem process of receiving and saving the raw range profile data has been
completed the embedded PC will send a flag to the Control Station PC. The Control
Station PC will move the platform to the next interval and the process repeats itself
automatically for the entire cross-range distance. It is further planned to have real-time
onboard data processing to provide better radiometric calibration accuracy as opposed
to offline post-processing.
The rail itself is of modular design, each being 1.5 meter in length. This enables
us to move the platform along any arbitrary distance, given a sufficiently flat length of
area. Several movement schemes can be chosen on the user interface. In the semi-
automatic scheme the user has to manually send the move and transmit commands on
the user interface. This is akin to a stop-and-go movement and is useful for
troubleshooting. The automatic with stops on interval scheme will move the platform
to each interval, stops on the interval location, and captures the range profile
automatically. The automatic continuous scheme moves and captures profiles while
moving. Figure 3.6 shows the Positioner Subsystem on-site experiment layout and
movement scheme. The RF front end, with the transmit and receive antennas mounted
on the mobile platform, is shown in Figure 3.7.
-
27
Figure 3.6: Positioner Subsystem layout and movement scheme.
Figure 3.7: The Positioner Rail SAR Imaging System mounted on modular rails used
on the test site.
Control
Station
Control
tether
Pla
tfo
rm m
ov
emen
t
Platform
Approximate
beam pattern
Modular rail
Beams from
successive pulses
Range direction
-
28
3.2 Ground test and measured data
A 50 meter long track was prepared on an open field with low background
clutter. A trihedral corner reflector was installed at a distance of 90 meters from the
track, one meter above the ground, to act as a point target. This arrangement is shown
in Figure 3.8. The side length of the trihedral should ideally be 10 times the signal
wavelength [43]. Working at 1.27 GHz the corner reflector should have a size of 2.5
meters on each side. This is prohibitively difficult thus a 1.80 meter trihedral corner
reflector was used instead. Our corner reflector was made using 0.6mm thick aluminum
sheets backed by 10mm thick plywood. This will result in a smaller than desired point
target reflection reading.
In this experiment the platform moves along the high precision track at 100 mm
per interval, which would yield roughly 450 sample points along the azimuth. The
semi-automatic movement scheme was employed. Captured data at each interval
would be stamped with its corresponding source location relative to the movement on
the track.
Figure 3.8: Arrangement of rail SAR transceiver and target corner reflector on test
site.
95m
Range length
Rail SAR
Transceiver
Corner reflector
50m
Azimuth
length
-
29
A series of field experiments have been conducted to verify the performance of
the CP-SAR system. These include point target measurement as mentioned in section
3.1.1, and ground-based SAR experiments using the developed moving platform. After
range compression, an ideal point target appears at each pixel of the recorded raw image
with the same amplitude. Azimuth range compression is achieved by multiplying the
azimuth range sample with the complex conjugate of the baseband reference signal at
frequency domain, and finally, the resultant signal is inverse Fourier transformed to
yield a focused image in both range and azimuth domain [44].
During the ground measurement, the system (antenna and sensor) was mounted
on the Positioner moving platform, facing perpendicularly to the identified natural
target. Radar echoes were captured in a stop-and-go manner with ground displacement
of 100 mm, along the 50 meter ground track.
The collected data was processed using a unique SAR signal processing
technique named Range Doppler Algorithm (RDA) [44]. RDA uses frequency domain
calculations in both range and azimuth. Each dimension is operated separately and in
the algorithm a range cell migration technique (RCMC) is performed in the range
Doppler domain. Figure 3.9 shows the point target return of trihedral at a distance of
92 meters.
Figure 3.9: Range measurement with point target return of trihedral detected at a
distance of 92 meters.
20 40 60 80 100 120 140 160 180 200
-80
-70
-60
-50
-40
-30
-20
-10
Range, m
Rela
tive P
ow
er,
dB
Range distance in meters
Rela
tiv
e p
ow
er in
dB
-
30
Figure 3.10: SAR Image obtained using datasets from 381 points.
The image obtained from this stop-and-go ground measurement is shown in
Figure 3.10. Incomplete focus of the point target seen in Figure 3.10 is due to total
collected datasets in the measurement is only 381 points and the measurement was
forced to stop due to bad weather conditions. The corner reflector could still be
sufficiently detected even though it is not at the optimum size. There is a fence along
the 60 meter mark which clearly is providing some unwanted reflections.
A new L-band ground-based SAR system operating at 1.27 GHz has been
designed and developed. The CP-SAR system can potentially provide high resolution
of SAR imagery for disaster monitoring. The Positioner moving platform that was
presented provided a stable test-bed to assess the L-band SAR imaging system. Setting
up the entire system does take considerable time due to large component sizes but the
benefit of this rail is providing a stable and accurate positioning for SAR ground
measurement and experimentation.
50 100 150 200 250
0
5
10
15
20
25
30
35
Range distance in meters
Cro
ss r
an
ge d
ista
nce
in
met
ers
-
31
Chapter 4 ANTENNA DESIGN
One of the most important subsystems of the SAR system is the antenna. Antenna
radiation patterns have side-lobes that add to ambiguity in the form of ghosting and
object repetition in SAR images. Applying filtering, such as the Hamming, to reduce
side-lobe level effects has a drawback of degrading image resolution. The aim of this
work is to develop an antenna suitable for the SAR system that has been designed with
a focus on side-lobe reduction using Dolph-Chebyshev synthesis and optimized feed
network. This chapter presents the development process of the proximity-coupled
corner-truncated patch array antenna. The proposed antenna was designed and then
simulated using the FEKO electromagnetic simulation tool before fabrication.
4.1 Reduced side-lobe Circularly Polarized Array Antenna
The radiation pattern design of the SAR antenna depends on factors such as
azimuth ambiguity, range ambiguity, range gain variation, and reflections from the
surrounding structures [47]. Range ambiguity is caused by echoes received from earlier
or later pulses arriving back to the antenna at the same time as the desired echo. Range
gain variation occurs due to the side-looking scan method. Correction is required to
compensate the difference of power delivered in range due to a large scan swath. Proper
design of the antenna fixture and suppression of the side-lobes can reduce reflection
significantly. SAR performance parameters were set and used to design the array
antenna. Microstrip patch antenna technology was chosen as it is lightweight and has
the potential to achieve the design requirements with proper design.
-
32
The proposed L-band proximity-coupled corner-truncated patch array antenna
is an iteration of our own previous design [45]. This array consists of 10 elements in a
2 x 5 arrangement. Each square radiating patch has a side length of 79 mm to operate
at the center frequency of 1.27 GHz. Circular polarization is achieved by truncated
corners. Two substrate layers are used for a total thickness of 3.2 mm; the top layer
consists of the radiating patches, the middle layer is the feed network which includes
T-junction power dividers. Quarter-wave transformers are employed to provide better
matching along the feed network. The feed line has a width of 5 mm to achieve an
impedance of 50 . The last copper layer is the ground plane. In this work we have
employed the NPC-H220A substrate from Nippon Pillar Corporation, having a
thickness of 1.6 mm, a relative permittivity of r = 2.17, and a loss tangent of = 0.0005.
4.1.1 Single element design
The first step is to design a single radiating patch, which will then be used as
elements in an array. A square patch with truncated corners is used. Changing the patch
size will change the center operating frequency. The feed point, determined by the
distance Fd, is offset from the center of the patch to provide optimum reflection
coefficient. Varying the truncation size will affect the axial ratio (AR), and thus the
level of circular polarization. Several samples made during the optimization process is
shown in Figure 4.1. By optimizing the corner truncation length, C, the axial ratio can
be minimized. When using the same design in an array, the suitable truncation size
requires re-optimization as its size depends greatly on the overall patch size, patch
spacing and feed network.
Figure 4.1: Effect of corner truncation size, C, on the axial ratio.
Frequency (GHz)
Ax
ial
Rat
io (
dB
)
0
2
4
6
8
10
1.23 1.24 1.25 1.26 1.27 1.28 1.29
C = 11 mm
C = 10 mm
C = 9.5 mm
C = 9 mm
C = 8 mm
-
33
4.1.2 Array design
An array with uniform power division typically results in good gain in the main
lobe, but also large side-lobe levels which can add to ambiguity. To further reduce the
signal-to-noise ambiguity, the side-lobes are reduced by utilizing the Dolph-Chebyshev
synthesis method.
The movement of the SAR sensor across the target area results in a Doppler
frequency shift proportional to the relative velocity of the target, and a time delay
proportional to the range distance. Due to the signal sampling rate, aliasing will occur
in the azimuth signals which results in ambiguity. Azimuth ambiguity is quantified with
the azimuth ambiguity-to-signal ratio (AASR) and can be calculated [45]. Based on the
AASR requirement and system PRF, a side-lobe level (SLL) of 20dB would provide
sufficient AASR for good SAR system performance. This value was used as input to
the Dolph-Chebyshev synthesis.
The other input for the synthesis is the distance between elements, d, which is
limited by the following equation:
0
1
max
1cos
zd
(4.1)
where is the wavelength at the operating frequency and z0 is the Chebyshev
polynomial matching variable [2]. Selecting a spacing larger than Eq. (4.1) will result
in unwanted grating lobes. The spacing is chosen to be 0/2, which complies with Eq.
(4.1). For the 2 x 5 element array, the array factor can be derived as:
uauaauAF 2cos2cos2)( 210 (4.2)
where
cos)/(2 du (4.3)
For d = 0 / 2, u = cos . By comparing Eq. (4.2) with the corresponding Chebyshev
polynomial, the normalized array factor equation becomes:
uuuAF 2cos)518.0(2cos)8327.0(21)( (4.4)
-
34
The antenna factor, AF(u), provides the theoretical beam pattern and is plotted as the
dotted line in Figure 4.2. The figure shows that the theoretical side-lobe level is 20 dB
below the maximum of the main lobe. Another value of importance in the figure is the
beamwidth that is 11 degrees. This design complies with the narrow azimuth
beamwidth requirement.
Table 4.1: Derived Dolph-Chebyshev Array Factors for 2x5 Planar Array
Patch designation Array Factor
Patch (1,1) and (1,2) 0.518
Patch (2,1) and (2,2) 0.8327
Patch (3,1) and (3,2) 1.0
Patch (4,1) and (4,2) 0.8327
Patch (5,1) and (5,2) 0.518
Figure 4.2: Theoretical, simulated and measured normalized antenna beam patterns of
the proposed array antenna.
4.1.3 Array signal distribution network
Once the array factors have been derived, these values can be used to design the
array signal distribution network. The distribution network will channel energy to each
patch at the right magnitude to generate the designed Dolph-Chebyshev beam pattern.
A corporate feed network was utilized with lengths kept short and line widths at a low
impedance. Most of the embedded microstrip line has an impedance of 35 while the
network has an input impedance of 50 and connected to an SMA connector. The
Angle, (degrees)
No
rmal
ized
gai
n (
dB
ic)
-40
-35
-30
-25
-20
-15
-10
-5
0
-90 -60 -30 0 30 60 90
Measured
EM Simulation
Theoretical20 dB
difference
-
35
proposed L-band proximity-coupled corner-truncated patch array antenna is shown in
Figure 4.3.
Unsymmetrical microstrip T-junctions are used in the feed network to divide
the power among the patches according to the array factor in Table 4.1. The geometrical
design of the feed is optimized using a method-of-moment based software. After
optimization, the array antenna designed shown in Figure 4.3 was fabricated with the
dimensions listed in Table 4.2. In total, four arrays were manufactured for measurement
and use in the SAR system; two units of left-handed circularly polarized (LHCP) and
two units of right-handed circularly polarized (RHCP).
(a) Top view
(b) Feed network, viewed from bottom (c) Side view
Figure 4.3: Dimensions of the proposed side-lobe reduced array antenna.
Table 4.2: Proposed Array Antenna Dimensions
Parameter Size (mm) Parameter Size (mm)
L 79.18 Sv 50.0
C 9.0 R 24.0
Fd 14.0 w1 6.8
F, Lh 40.0 w2 5.0
Lv 38.0 Lground 655.0
Sh 54.0 Wground 285.0
Sv
Sh
C
L
L x
y
z
Lground
Wg
roun
d
L
h
Fd
Lv
w2
w1
F
x
y
z Ground
z
yx
Bottom substrateTop substrate
Feed layer
Patch
-
36
4.1.4 Simulated and Measured Results
A fabricated LHCP array antenna is shown in Figure 4.4. The antennas
reflection coefficient and radiation pattern were measured in an anechoic chamber and
compared to simulated performance. In Figure 4.5, the reflection coefficient shows
good agreement between simulation and measured values. The impedance bandwidth
achieved is 58 MHz with a minimum reflection coefficient of -20 dB at around the
center operating frequency. At maximum bandwidth, the antennas can function at
frequencies from 1.241 to 1.299 GHz. The realized center frequency is slightly higher
most probably due to errors in fabrication of the radiating patch size.
(a) Radiating layer
(b) Feed network layer
Figure 4.4: The top and bottom layers of the LHCP fabricated array.
Figure 4.5: Simulated and measured reflection coefficient of the RHCP array antenna.
-25
-20
-15
-10
-5
0
1.1 1.15 1.2 1.25 1.3 1.35 1.4 1.45
Measured
Simulation
Frequency (GHz)
Refl
ecti
on
co
eff
icie
nt
(dB
)
Desired minimum level
Bandwidth = 58 MHz
-
37
The radiation pattern in Figure 4.2 shows the measured side-lobe level to the
left of the main lobe to be 20 dB. This matches the theoretical and simulated levels.
However on the right of the main lobe the side-lobe levels are not as low. This can be
attributed to the feed design. From the feed designs seen in Figure 4.3(a) and Figure
4.5(b), there are more microstrip lines on the right side. These will contribute to
spurious radiation. Figure 4.6 shows the lopsided beam pattern on a polar graph.
Figure 4.6: Simulated and measured radiation pattern of the RHCP array antenna.
Figure 4.7: RHCP array antenna simulated and measured axial ratio and gain against
frequency.
-45
-40
-35
-30
-25
-20
-15
-10
-5
00
30
60
90
120
150
180
210
240
270
300
330
Measured without casing
EM Simulation
0
2
4
6
8
10
12
14
0
1
2
3
4
5
6
7
8
9
1.22 1.24 1.26 1.28 1.3
Measured
Simulated
Frequency (GHz)
Ax
ial
Rat
io (
dB
)
Gai
n (
dB
ic)
-
38
Figure 4.8: RHCP array antenna simulated and measured axial ratio against angle.
The simulated and measured AR and gain against frequency is plotted in Figure
4.7, showing the measured gain is 10.011.9 dBic from 1.241 to 1.299 GHz. The 3-dB
AR bandwidth achieved at the direction of = 0 (i.e., the AUT is set perpendicular to
the standard antenna during parameter measurement) is about 17 MHz, which
corresponds to 1.36% of the operation frequency of 1.27 GHz. In the simulation, on the
other hand, the value is 13 MHz, or around 1.00% of the operation frequency. The
minimum axial ratio of the measured data is around 1.40 dB, exhibiting a difference of
around 0.64 dB as compared with the simulation result. Inaccuracies in the
implementation of the corner-truncated patches are one possible cause of differences
between the simulated and measured results. As explained in Section 4.1.1, the
truncation size greatly effects the axial ratio. Displaying AR against angle in Figure 4.8
shows that the designed array antenna has a measured axial ratio of
-
39
4.2 Beam tilted Circularly Polarized Array Antenna
In this section, a circularly polarized antenna array employing beam tilting as
well as side-lobe reduction using the Dolph-Chebyshev synthesis method at C-band 5.3
GHz is presented. The antenna developed in this work is to be used in an airborne SAR
system. Comparatively to spaceborne systems, airborne SAR operates at a much lower
altitude and thus has a different set of operating parameters such as shorter range, and
a lower average transmit power. In our case the SAR range is around 15 km with an
average transmit power of around 50 watts based on a p